rp7420 stopped reboot and MP can't access

[Shirley Zhang]

I have rp7420 with only one partition and before it finished rebooting, it hang and MP can't access either. I switch the slave MP and from the log it shows "Warning
Keyword: MASTER_MP_FAILURE
The mp in the master slot of the specified cabinet is reporting failure"; now even I can access MP but it can't detect any boot devices. My question is : how can I make the server work? (MP>CM>PS shows all hardware OK.

Thanks for your input.

Shirley

[Torsten.]

Hi Shirley,

first of all, unplug the box for some minutes and try again. Second: be sure to have a current firmware on all components.
After the power cycle, do a "sysrev" on GSP level and post it or give HP a call.

[Adisuria Wangsadinata_1]

Hi,

You can use DF from MP prompt to check which part is having a problem.

Check this document (docID : KBQA00000153) for your reference, about Superdome/Keystone BCH/MP command menu :

http://www2.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000080074590

Call HP for this issue.

Hope this information can help you.

Cheers,
AW

[Shirley Zhang]

Hi,

My big question is if one core I/O board fails, can server work with the redundant core I/O? My case is that when server boots, VFP shows Ok(cell joins the partition), but boot stops at the BCH mode because running seach returns :
No Devices Found, even I insert an CDROM it still returns " No Devices Found".

Thanks,

Shirley

[Silvino]

Hi,
Core I/O are not redundant on RP7420.
You have 1 core i/o for each cell board.
core i/o for cell 0 has hw path 0/0/8/1 (slot8 right)
Core i/o for cell 1 has hw path 1/0/1/1. (slot1 left)

The 2 bottom disk assigned to MP0(ids 5 and 6)
Upper left is assigned to MP1
Upper roght is assigned to core i/o for cell 1.

If you can not "see" the disks with "SEA", you may have a failed MP (or core i/o) and you should call HP to repair/replace the MP.
Rgrds,
Silvino

[Shirley Zhang]

Thanks Silivino,

I have one MP failed for sure, but the other MP works, can I only use one MP to reboot rp7420 or I must have two MPs?

Shirley

[Sameer_Nirmal]

rp7420 provides redundancy at MP level if it has two MPs installed in single npar configuration. Two MPs work in primary ( master ) and secondary ( slave ) modes.
If the master fails the secodary should take over through failover process.

By default MP1 which belongs to chassis 1 acts as primary MP. I believe it is failed in your case. You can check the status of the failover from seconday MP's menu
MP->CM->PS and M->CM->DE options.
It seems the failover has not taken place in the server. You can force the failover to the secodary MP using MP->CM->XD 9 option.

[Mohanasundaram_1]

HI Shirley,

Usually the configuration is in such a way that the root disk and its mirror on different Core I/O SCSI.

This will enable to boot the system with
hpux -lq
in case of one disk or SCSI controller failure.

It will be important to know how your is system configured. If there is no mirror disk then the system is not going to boot on this failure.

Otherwise, try to boot in the alternate boot path without checking the LVM quorum.

With regards,
Mohan.

[Duncan Edmonstone · Accepted solution]

Ummm I'm afraid that MP failover is *not* supported any more - see the following:

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=c00462190

Sorry to be the bearer of bad tidings

HTH

Duncan

[Stefan Stechemesser]

Hi,

Duncan is right, MP failover is no longer supported.

Anyway, in your situation it has already happened.

Regarding the boot problem:

The "sea" command only search for bootable devices on the "CoreCell", the cell with the Core I/O.

Try "sea all" to search on both cells.

In your case, it may be that the complete Core I/O 1 has failed (not only the MP part) and therfore, your Core Cell is now 0 and no longer 1.
You can check this with the "corecell" command in the Configuration Menu of BCH.
With the "io" command in the information menu, you can see if the Master core I/O is still be seen by the partition (hw path 1/0/0/3). If not, then the disks in the upper left slot (1/0/0/3/0.6) will also no longer be accessible.
The disks in the other slots are accessed by the Procorium Lan/SCSI card and the other Core I/O.

best regards

Stefan